Write introduction section paragraphs for academic papers using wave-based execution. Triggers when /aw-execute assigns a Wave 1 or Wave 2 task for the introduction section (1.1, 1.2, 1.3, 1.4). Reads research-brief.json, literature.md, and methodology.md to produce independent paragraph .tex files. Each task writes one paragraph file following Elsevier LaTeX format.
Write introduction section paragraphs as independent wave tasks. Each invocation writes one paragraph file following the ROADMAP task decomposition. Paragraphs are written in parallel when assigned to the same wave (no file overlap), and sequentially when dependencies require it.
/aw-execute assigns a wave task with section: introduction1-1, 1-2, 1-3, 1-4.planning/research-brief.json — Problem statement, novelty claims, contribution bullets, author intent.planning/literature.md.planning/methodology.md — Technical approach overview for contribution framingtemplates/elsevier/sections/introduction.tex — LaTeX format reference (Elsevier template)manuscripts/{paper-name}/sections/intro/
├── 1-1-background.tex # \paragraph{1.1 Research Background}
├── 1-2-problem.tex # \paragraph{1.2 Problem Definition}
├── 1-3-contributions.tex # \paragraph{1.3 Main Contributions}
└── 1-4-structure.tex # \paragraph{1.4 Paper Structure}
Each file contains exactly one \paragraph{} block with appropriate content.
Input context: Problem domain, broad research area, recent developments
Output: 2-3 paragraphs establishing the research context
LaTeX structure:
paragraph{Reviewing Past Work}
\label{sec:intro:background}
[2-3 paragraphs tracing the evolution of the field,
citing key prior work, narrowing down to the specific
sub-area the paper addresses]
Coverage:
Input context: Specific gap or limitation identified in research-brief.json
Output: 1 paragraph precisely defining the problem
LaTeX structure:
paragraph{Problem Statement}
\label{sec:intro:problem}
[1 paragraph clearly stating the specific problem,
limitation, or gap that motivates the paper.
Avoid vague language; be precise about what is missing
or suboptimal in existing work.]
Coverage:
Input context: Novelty claims and contribution bullets from research-brief.json
Output: 3-4 bullet items summarizing contributions
LaTeX structure:
paragraph{Main Contributions}
\label{sec:intro:contributions}
\begin{itemize}
\item We propose \ldots
\item We demonstrate \ldots
\item We provide \ldots
\end{itemize}
Coverage:
Input context: Section ordering from ROADMAP.md
Output: 1 brief paragraph outlining paper structure
LaTeX structure:
paragraph{Paper Organization}
\label{sec:intro:structure}
The remainder of this paper is organized as follows.
Section~\ref{sec:methodology} describes \ldots
Section~\ref{sec:results} presents \ldots
Section~\ref{sec:discussion} discusses \ldots
Section~\ref{sec:conclusion} concludes \ldots
Coverage:
paragraph{Title}
\label{sec:intro:task-id}
Content here...
% For contributions, use itemize inside the paragraph block:
\begin{itemize}
\item First contribution...
\item Second contribution...
\end{itemize}
\cite{key1,key2}\citealp{key1,key2,key3}\citeauthor{smith2020} found that...From templates/elsevier/main.tex:
\documentclass[review]{elsarticle}
\usepackage{amsmath,amsfonts,amssymb}
\usepackage{graphicx}
\usepackage{cite}
\label{sec:intro:background}\label{sec:intro:problem}\label{sec:intro:contributions}\label{sec:intro:structure}1-1, 1-2, etc.)manuscripts/{paper-name}/sections/intro/ if not exists\paragraph{} block and labelFile: manuscripts/my-paper/sections/intro/1-1-background.tex
paragraph{Reviewing Past Work}
\label{sec:intro:background}
Deep learning has transformed computer vision over the past decade, with
convolutional neural networks achieving human-level performance on benchmark
datasets \cite{lecun2015deep,krizhevsky2012imagenet}. Recent advances in
transformer architectures have further improved results on complex visual
understanding tasks \cite{dosovitskiy2021image,vit2020}. However, despite
these improvements, existing approaches face challenges when applied to
low-resolution medical imaging data \cite{smith2021limitations}. In the
domain of remote sensing, similar limitations have been observed, where
computational costs grow prohibitive for large-scale deployment \cite{chen2022remote}.
The field of low-light image enhancement has seen particular interest,
with traditional histogram equalization methods \cite{pizer1987adaptive}
gradually replaced by learning-based approaches \cite{lorenz2021learning}.
Despite these advances, real-time enhancement for video streams remains
computationally challenging \cite{wang2023real}, motivating the need for
more efficient architectures suitable for edge deployment.
This skill is spawned by aw-execute (wave executor) for each introduction paragraph task. It is not triggered directly by the user. The wave executor merges all paragraph outputs into sections/introduction.tex after all wave tasks complete.
aw-execute assigns:
Task: 1-1 | Section: Introduction | Paragraph: Research Background
→ Spawns aw-write-intro for 1-1-background.tex
→ Writes: sections/intro/1-1-background.tex